Mech. Eng. Design of Synchrotron Radiat. Equip. and Instrum.

MEDSI2020, Chicago, IL, USA JACoW Publishing

Content from this work may be used under the terms of the CC BY 3.0 licence (© 2021). Any distribution of this work must maintain attribution to the author(s), title of the work, publisher, and DOI
ISBN: 978-3-95450-229-5 ISSN: 2673-5520 doi:10.18429/JACoW-MEDSI2020-TUPC14

COPPER BRAID HEAT CONDUCTORS

FOR SIRIUS CRYOGENIC X-RAY OPTICS
F. R. Lena†, G. V. Claudiano, J. C. Corsaletti, R. R. Geraldes, D. Y. Kakizaki, R. L. Parise,
M. Saveri Silva, M. S. Souza, L. M. Volpe, LNLS, Campinas, Brazil

Abstract copper thermal conduction at low temperatures has a

strong correlation with its electrical conductivity [2] and is
The low emittance and high photon flux beam present at
heavily affected by impurities as oxygen and sulphur, lead-
the 4th-generation Sirius synchrotron light source beam-
ing to the correct material selection [3].
lines result in high energy densities and high heat loads at
some specific X-ray optics such as monochromators and Commercial and 1st-Generation In-House
white beam mirrors. This challenges the design of such sys- Braids
tems since the introduction of thermal stresses may lead to
optical surface deformation and beam degradation. Thus, An initial effort of in-house development of copper
to keep the systems within acceptable deformations some braids for mirrors and monochromators was made between
of the optical elements are cryogenically cooled. However, the years of 2017 and 2019. The first systems were de-
this poses the requirements of decoupling the thermal sinks signed by the LNLS team and manufactures by a local part-
(cryostats) from the optics and the mechanisms to maintain ner (Barbanera Qualità) and consisted of pressing multiple
their desired degrees of freedom for alignment and dy- braids of 2.5mm in diameter inside electrical-discharge-
namic operation. In this context we present the develop- machined (EDM) cavities in the electrolytic copper end-
ment of low-stiffness copper-braid-based heat conductors, blocks. The braids were made of 588 40 AWG stranded
summarizing the motivation and main aspects regarding electrolytic-tough-pitch (ETP) copper wires by the national
their fabrication and application at the beamlines. manufacturer Indel LTDA. The pressing was performed
with a 100-ton hydraulic press in an open-die fashion, i.e.
INTRODUCTION without any mold. The compaction criteria were visual,
with the pressing process being stopped when there were
For high heat-loads monocrystalline silicon optics at Sir-
no visible gaps between the wires and the block. The large
ius beamlines, one of the standard design concepts for
end-blocks were then machined, and the entire braid sys-
beam-load deformation suppression is the use of liquid ni-
tem gold plated to reduce radiation heat transfer.
trogen (LN2) cryostats for cooling the elements down to
Despite the good geometry and low-stiffness results,
near 125K, where their coefficient of thermal expansion is
they presented much lower thermal conductivity when
virtually zero and deformations due to thermal gradients
compared to commercial systems that were already in use.
are minimized [1]. In this context, low-stiffness copper
In fact, some of the braided systems had a tested perfor-
braids are widely used to thermally couple the optics to the
mance as low as 57% of the theoretical calculated thermal
cryostats while limiting the mechanical coupling between
conductivity (CATERETÊ beamline M2 braid). Initial in-
them, such that external vibration disturbances are avoided
vestigations performed via microscopy, after cutting both a
and kinematic is preserved. Even though commercial solu-
commercial braid and an in-house one (Fig. 1), highlighted
tions for cryogenic, ultra-high vacuum (UHV) compatible
the potential culprit, namely: the poor wire compaction in-
copper braids exist worldwide, the high costs, long lead
side the end blocks. It was found that the cross section of
time and customization limitations stimulated the develop-
the first one had an average void density of 4.2%, whereas
ment of an in-house solution with local partners. For spe-
the problematic one had 24.1% (ImageJ). Also, the wires
cial applications, either because of complex geometry or
were much less deformed, leading to limited contact area
heat extraction capacity, smaller braided modules can be
between them due to insufficient pressing forces. This re-
soldered into larger systems using low temperature fillers.
sulted in a poor thermal coupling among both the wires
themselves and the end-blocks.
COPPER BRAID MANUFACTURE
The basic manufacturing process consists in cold forging
(ambient temperature), or pressing, copper ropes inside
bulky copper terminals. The copper ropes are usually made
from cold-drawn thin wires stranded together to form a
small braid with the diameter in the millimeter range. This
results in an all-copper, weld free braid system with solid
ends that can be machined for different geometries. As Figure 1: Micrographies of the commercial (left) and the
there is no metallurgical joint between the wires them- 1st-generation in-house (right) solutions from section-cut
selves and the end-blocks, the heat must be carried across of the braids inside the end-blocks.
multiple contact interfaces, making the correct design of The hypotheses for the bad pressing consisted of a com-
these interfaces a point of attention. Furthermore, the bination of insufficient hydraulic pressing forces and ex-
___________________________________________

† francesco.lena@lnls.br cessively large and stiff end-blocks, resulting in low plastic

Core technology developments TUPC14
Cryogenics 207
 Mech. Eng. Design of Synchrotron Radiat. Equip. and Instrum.
Content from this work may be used under the terms of the CC BY 3.0 licence (© 2021). Any distribution of this work must maintain attribution to the author(s), title of the work, publisher, and DOI
ISBN: 978-3-95450-229-5 ISSN: 2673-5520
deformation. These results led to new efforts being made
towards improving the manufacturing process, focusing on
better compaction and deformation of the wires.
2nd-Generation In-House Braids
To reduce the end-block stiffness for better pressing, we
abandoned the concept of starting from the large copper
blocks for subsequent machining. Instead, those termina-
tions were standardized as 1x1x1/4’’ (LxWxH) for systems
with a single row of 9 of the same 2.5mm ETP braids, and
1x2x1/4’’ for the single-row 18-braid ones. In this manner,
most of the termination internal volume was comprised by
the wires. After the pressing, the blocks were fine ma-
chined to remove the extra material (approx. 1 mm) and for
hole drilling. Also, the blocks were manufactured out of
completely annealed OFHC copper, increasing the load
transfer to the wires due to their low yield strength. The
pressing strategy and load were varied aiming the lowest
void density among wires.
Open Versus Closed Die Pressing
Initial tests with the new 9 braids, 1x1’’ geometry was
performed without any die and a pressing load of 30 Ton
(456 MPa). While already achieving a better pressing per-
formance, as compared to the 1st-generation systems, the
open fashion resulted in an anisotropic wire load distribu-
tions and highly deformed end-block, making it difficult to
machine the final form. Also, the load could not be further
increased without breaking the blocks.
To better distribute the pressing loads to the wires and
blocks, a U-shaped, 2-part die was machined out of high-
speed steel (HSS) with the exact final geometry of the de-
sired end-blocks. Then, new tests were performed with the
same load. The void densities achieved were 6.6 and 6.8%
for the open and closed die, respectively.
Pressing Load
A new U-shaped HSS closed die was made to admit the
larger 1x2”, 18-braid system. The pressing was then redone
with the same 456 MPa compression load (60 tons scaled
total force), reaching a slight higher void density of 8.2%.
Following this, an even higher load of 570 MPa (75 tons
total force) was used to press another 18-braid system. This
time, the final average void density was of only 1.7% and
was considered enough as it was already better than the
well performing commercial braids (4.2%) (see Fig. 2).
Figure 2: Closed die pressing with different loads.
Conditioning for UHV
To remove contaminants introduced during manufactur-
ing for the UHV application compatibility, the systems
TUPC14
208
MEDSI2020, Chicago, IL, USA JACoW Publishing
doi:10.18429/JACoW-MEDSI2020-TUPC14
were sonicated with: IC115 + deionized (DI) water solution
(1:4, volume); then, pure DI water; and, finally, pure iso-
propanol, 30 minutes each. After that, cleansing was made
in a sulphuric acid and hydrogen peroxide (1:1, volume)
bath, aka. Piranha solution, for approximately 2 minutes,
followed by DI water rinse and isopropanol sonication for
15 minutes. The braids were, then, vacuum backed at
120°C for 24h, and stored in vacuum until assembled or
gold platted. Residual gas analyses (RGA) were performed
with a Stanford SRS RGA200 up to 200 Da in a ~1e-6 mbar
vacuum, with no specific contaminants detected. Indeed,
some systems are currently working at < 5e-10 mbar. In the
cases of gold platting, the braids were still re-cleaned with
the same IC115/DI water solution, followed by isopropanol
sonication for 30 minutes and the same baking procedure.
THERMAL CONDUCTION TESTS
To validate the thermal conductivity of the braids, a ded-
icated setup was developed (Fig. 3). It consisted of an alu-
minium machined bar coupled to a Janis ST400 LN2 cryo-
stat, to which one of the end-blocks of the braids was cou-
pled. 2k Ohm IST resistive temperature detectors (RTD)
were attached to both braid end-blocks and read by a NI
cRIO 9226 card. Thin film Kapton heaters were attached
to the other extremity and driven by a Keithley controller.
Indium foil was used in all interfaces. The entire system
was kept under high vacuum (< 1e-6 mbar).
Figure 3: In-vacuum cryogenic thermal test setup.
After cooling, the temperature offset of both termina-
tions caused by the radiation heat transfer was normalized
and power was delivered to the heaters in controlled steps.
The thermal conductivity was then calculated by the simple
division of the total power delivered to the heaters by the
temperature difference of both end-blocks, with correc-
tions for cable losses and radiation heat transfer. Some of
the results are shown at Table 1. For all tests, the deter-
mined conduction efficiency varied between 82 and 91%
of the theoretical values.
Table 1: Thermal Conduction Test Results
Braid ID Braid Length Mean Power Conductivity
n# (mm) Temp. (K) (W) (W/K)
SCB1806 18 60 111.4 5.55 0.314
SCB1806 18 60 131.0 9.78 0.293
SCB0910 9 100 111.0 2.35 0.101
SCB0910 9 100 139.6 5.25 0.094
Core technology developments
Cryogenics
Mech. Eng. Design of Synchrotron Radiat. Equip. and Instrum. MEDSI2020, Chicago, IL, USA JACoW Publishing

Content from this work may be used under the terms of the CC BY 3.0 licence (© 2021). Any distribution of this work must maintain attribution to the author(s), title of the work, publisher, and DOI
ISBN: 978-3-95450-229-5 ISSN: 2673-5520 doi:10.18429/JACoW-MEDSI2020-TUPC14

SOLDERING AND BRAZING BEAMLINE SYSTEMS

A common fashion for thermally coupling solid bodies The CARNAÚBA beamline M1 mirror along with an as-
consists in layering a soft, good conductor interface mate- sociated diffraction beam diagnostic is currently the cryo-
rial and compressing it with a static load. Due to the need genic system relying on copper braids with the highest
for UHV compatibility and cryogenic properties, annealed power load at Sirius [6]. Indeed, the total power in the com-
indium foil (0.2 mm, Goodfellow) is usually employed be- missioning phase of the storage ring is approximately 77
tween the conduction blocks, mirrors, and cryostats at Sir- W [7], but it should reach 100 W in the near future.
ius. An acceptable thermal conductivity value for this solu- To solve the needed cooling requirements, while pre-
tion currently used in many Sirius thermal models is 3 serving the optics alignment and motion, a set of 12x 18-
kW/m2K, based on [4]. Nevertheless, since the new devel- rope braids were used to couple a cryostat-cooled copper
oped braids end-block geometries are standardized, they block to the optics interface sets (Fig. 5). All copper sur-
often must be coupled to intermediary parts to form the faces were gold plated to reduce radiation heat exchange.
thermal links, but any new interface may lead to a loss in Solid indium foil was used as interface material in all cases
thermal efficiency. To improve the interface conductivity but will be indium-soldered soon. The system has been
in relation to the pressed indium, soldering and brazing proven fully functional during the beamline commission-
tests were performed in vacuum (<1e-5 mbar) with multi- ing.
ple fillers (Table 2).
Table 2: Brazing and Soldering Tests
Filler FillerThickness Temp. Time Category
Material (mm) (°C) (min)
Palcusil 10 0,1 885 5 Brazing
Cusil 0,1 780 5 Brazing
Sn100C 0,3 227 30 Soldering
Sn100CV 0,3 227 30 Soldering
Indium 0,5 157 30 Soldering
Figure 5: Braided system at the CARNAÚBA beamline
Figure 4 shows the micrographies for the several cases. M1 mirror and diagnostic system.
Despite both Cusil and Palcusil resulted in thinner, metal- Also, both the M2 mirror and SSA (secondary source ap-
lurgically bonded interfaces, the high temperatures em- erture [8]) at CARNAÚBA use the same developed braids,
ployed during the process caused an undesired adhesion of but have already indium-soldered interfaces, as they were
the copper wires among themselves by diffusion bonding, used as experimental systems, having the best performance
as well as strong annealing. This led to a completely loss so far. Other systems, as the CARNAÚBA 4CM mono-
of flexibility in the braids, as the wires could no longer slip chromator [9] and IPÊ M4 and M5 mirrors, also use these
among themselves, invalidating the application. Thus, new braids, and many others are on the way.
efforts for low temperature bonding (soldering) using tin
based fillers, namely, Sn100C and Sn100CV, and pure in- CONCLUSION
dium were made.
A manufacturing process for fabricating weld-free, low-
cost, high-efficiency flexible copper braids for cryogenic
applications was developed in-house with the help of a lo-
cal partner. The main challenge of achieving a high com-
paction at the braids solid blocks was solved with: a closed
die pressing process, optimized end-block geometries, and
copper annealing for lower yield strength, resulting in a
void density as low as 1.7% when cold pressed with 570
MPa. Both the vacuum and thermal performances were
tested with successful results, with systems working down
to < 5e-10 mbar and conduction efficiency between 82 and
Figure 4: Different brazing and soldering interfaces be- 91% of the theoretical values. Large optical systems cur-
tween copper end-blocks. rently use the developed systems at Sirius up to 100 W.
Although the Sn100C and Sn100CV fillers resulted in
even, reproducible interfaces, it was later found that both
ACKNOWLEDGEMENTS
alloys contained a considerable amount of phosphorus The authors would like to gratefully acknowledge the
(0.73 and 0.68% w.t. respectively, measured by inductively Brazilian Ministry of Science and Technology for the fund-
coupled plasma mass spectrometry), being unsuitable for ing, as well as all the engineering and science team mem-
vacuum applications near optical surfaces. Therefore, the bers working for the Sirius project. A special thanks also
pure indium soldering method was chosen, having an ex- goes to the people from Barbanera Qualità for all the shared
pected conductivity near 400 kW/m2K [5]. efforts towards developing the braids solution.
Core technology developments TUPC14
Cryogenics 209
 Mech. Eng. Design of Synchrotron Radiat. Equip. and Instrum. MEDSI2020, Chicago, IL, USA JACoW Publishing

Content from this work may be used under the terms of the CC BY 3.0 licence (© 2021). Any distribution of this work must maintain attribution to the author(s), title of the work, publisher, and DOI
ISBN: 978-3-95450-229-5 ISSN: 2673-5520 doi:10.18429/JACoW-MEDSI2020-TUPC14

REFERENCES
[1] R. R. Geraldes et al., “The Design of Exactly-Constrained X-
Ray Mirror Systems for Sirius”, in Proc. 10th Mechanical
Engineering Design of Synchrotron Radiation Equipment and
Instrumentation Int. Conf. (MEDSI'18), Paris, France, Jun.
2018, pp. 173-178. doi:10.18429/JACoW-MEDSI2018-
WEOAMA04
[2] G. White, “The Thermal and Electrical Conductivity of Cop-
per at Low Temperatures,” Aust. J. Phys., vol. 6, no. 4, 1953.
doi:10.1071/ph530397
[3] S. S. Rosenblum, W. A. Steyert, and F. R. Fickett, “A simple
method for producing high conductivity copper for low tem-
perature applications,” Cryogenics, vol. 17, no. 11, 1977.
doi:10.1016/0011-2275(77)90123-0
[4] D. A. Schneider, “Thermal Contact Resistances in a Thermal
Conductivity Test System”, Master Thesis, Oregon State Uni-
versity, 1998.
[5] R. Radebaugh, “Thermal conductance of indium solder joints
at low temperatures,” Rev. Sci. Instrum., vol. 48, no. 1, 1977.
doi:10.1063/1.1134856
[6] V.B. Zilli et al., “Installation and Commissioning of the Ex-
actly-constrained X-ray Mirror Systems for Sirius/LNLS,”
presented at MEDSI 2020, Chicago, USA, July 2021, paper
MOPB06, this conference.
[7] L. M. Volpe, B. A. Francisco, S. M. Saveri Silva, J. C.
Corsaletti, R. R. Geraldes, “Thermal Model Validation for the
Cryogenic Optical Systems for Sirius/LNLS,” presented at
MEDSI 2020, Chicago, USA, July 2021, paper WEPC07, this
conference.
[8] H. C. N. Tolentino et al., “Innovative instruments based on
cryogenically cooled silicon crystals for the CARNAÚBA
beamline at Sirius-LNLS,” in AIP Conference Proceedings,
vol. 2054, p. 060026, 2019. doi: 10.1063/1.5084657
[9] M. Saveri Silva et al., “Four-Bounce Crystal Monochroma-
tors for the Sirius/LNLS beamlines,” presented at MEDSI
2020, Chicago, USA, July 201, paper MOPB04, this confer-
ence.

TUPC14 Core technology developments

210 Cryogenics